Trent Kyono, The Boeing Company; Jacob Lucas, The Boeing Company; Michael Werth, The Boeing Company; Justin Fletcher,Qdyssey Systems Consulting; Ian McQuaid, AFRL
Keywords: Deep learning, machine learning, multi-frame blind deconvolution
Abstract:
Astronomical images collected by ground-based telescopes suffer from degradation and perturbations attributed to atmospheric turbulence. The Multi-Frame Blind Deconvolution (MFBD) algorithms that can extract well-resolved images from these degraded data frames are computationally expensive, requiring supercomputing infrastructure for relatively fast performance (on the order of hours to resolve a LEO pass) and currently cant be done in real-time. Because of this, optimal collection parameters cannot be adjusted for maximizing the likelihood of producing a resolved image with MFBD. In this paper we present a neural network that predicts whether an MFBD algorithm will be able to resolve a degraded image (or sequence of images) in real-time, and present experiments conducted on actual data collected at the Maui Space Surveillance Site.
Date of Conference: September 17-20, 2019
Track: Machine Learning for SSA Applications